Hypoxia-Selective Antitumor Agents. 7. Metal Complexes of Aliphatic

of activated aliphatic mustards which can act as diffusible cytotoxins. Two series of Co(II1) complexes of the bidentate bisalkylating nitrogen mustar...
0 downloads 0 Views 1MB Size
Download PDF
J.Med. Chem. 1993,36, 1839-1846

1859

Hypoxia-Selective Antitumor Agents. 7. Metal Complexes of Aliphatic Mustards as a New Class of Hypoxia-Selective Cytotoxins. Synthesis and Evaluation of Cobalt(II1) Complexes of Bidentate Mustards David C. Ware,*Jpf Brian D. Palmer,' William R. Wilson,: and William A. Denny'*? Cancer Research Laboratory and Section of Oncology, Department of Pathology, School of Medicine, and Department of Chemistry, University of Auckland, Private Bag 92019, Auckland, New Zealand Received January 25, 1993

Nitrogen mustards coordinated to Co(II1) are potential hypoxia-selective cytotoxins, since oneelectron reduction to the Co(I1) complexes greatly labilizes the Co-N bonds, causing the release of activated aliphatic mustards which can act as diffusible cytotoxins. Two series of Co(II1) complexes of the bidentate bisalkylating nitrogen mustard ligands N,"-bis(2-chloroethyl)ethylenediamine (BCE) and Nfl-bis(2-chloroethy1)ethylenediamine (DCE) have been synthesized and evaluated for their hypoxia-selective cytotoxicity against AA8 cells in uitro. The complexes (acac) auxiliary ligands; cyclic voltammetry studies show also bear two 3-alkylpentane-2,Cdionato that variation of the alkyl group in the auxiliary ligands alters the reduction potentials of the complexes(within a series) over a range of about 150mV. In both series, the patterns of cytotoxicities of the cobalt complexes were broadly similar to those of the respective free ligands, suggesting that the cytotoxicity of these compounds is due to release of the free ligands. The nonsymmetrical ligand DCE and its cobalt complexes were 1 order of magnitude more cytotoxic than the corresponding BCE compounds. Although the unsubstituted acac/DCE complex showedno hypoxic selectivity against repair-deficient UV4 cells in a stirred suspension culture assay, the methyl and ethyl analoguesshowed substantialselectivity. The results may indicate anarrow range of acceptable reduction potential, with an optimum close to that for the methyl analogue ( E I ~=z-305 mV). The methyl analogue also shows hypoxic selectivity against repair-proficient cell lines (e.g., AA8 and EMT6) and has high activity against EMT6 cells in intact spheroids, suggesting that the released DCE is capable of back-diffusion from the hypoxic core of the spheroid. This work shows that metal complexes of nitrogen mustards have significant hypoxia-selective cytotoxicity toward mammalian cells in cell culture and are a new general class of hypoxia-selective cytotoxins. Hypoxic cells in solid tumors are resistant to ionizing radiation, and in some cases limit the ability of radiotherapy to control tumors.lV2 The noncycling status of many hypoxic cells3 and the difficulty of achieving adequate drug concentrations in regions distant from functional blood vessels4suggests that hypoxic cells may also often be resistant to conventional chemotherapeutic agents. Drugs which are selectively cytotoxic under hypoxic conditions are thus of potential value in cancer chemotherapyand radiotherapy. A particular attraction of this approach is that, because normal tissues are well oxygenated, hypoxia-selective cytotoxins (HSC) are expected to have selectivity for tumors. Although hypoxic cells are only a small subpopulation in most tumors: it has been argued that repeated administration of HSC might turn hypoxia to advantage by killing tumor cells as they cycle through a hypoxic states6 In addition, if bioactivation of drugs in hypoxic regions generates relatively stable cytotoxins capable of limited diffusion, then tumor hypoxiacould be exploitedto kill surrounding tumor cells at higher oxygen concentration~.~t~ Three classes of drugs (nitro aromatics (11, quinones, and N-oxides) are known to be activated by reduction via metabolic pathways which are inhibited by 0xygen.I In each case, selectivity for hypoxic cells is achieved by backoxidation of the initial one-electron adduct (e.g., nitro aromatic (2) or semiquinone radical anion) by oxygen, thus suppressing metabolic activation in oxygenated cells + Cancer Research Laboratory. t Section of Oncology. 1 Department of Chemistry.

Scheme I (a)

Ar-l;r02

-.s;t Ar;NOf

-

-+ Ar-NO

-c

+ Ar-NH2

02 .02

(b)

[cO1ll@

3

m 02 .02

[c011&]2+ -!% 4

[cOl'(H20)6]2+ + 6 L 5

(e.g., Scheme Ia). However, in many cases activation can also be effected by two-electron reduction (e.g., to the hydroquinone or nitroso compound) thus bypassing the oxygen-reversible intermediate. NAD(P)Hquinone oxidoreductase (DT diaphorase) is a well-known example of a two-electronreductase which can activate quinones9and, with lower efficiency, nitro cornpounds'OJ1 under aerobic conditions. Other reductases such as xanthine dehydrogenase also appear to have some ability to reduce quinone@ and nitroheterocycleslSunder aerobic conditions by two-electron reduction. An attractive alternative chemistry for hypoxia-selective metabolism is the use of complexes of transition metals for which only a one-electron reduction is possible." Cobalt complexes containingnitrogen mustard ligands are of particular interest. The cytotoxicity of nitrogen mustards depends on the electron density on the mustard nitrogen,I6 which controls ita alkylating reactivity. Coordination of the nitrogen lone pair to Co(II1) should suppress its toxicity since the electron pair is no longer availableto act as anucleophile. The de low-spinelectronic configuration of octahedral Co(II1)complexes (3)renders them kinetically inert (for example, the half-life for the aquation of [Com(NHs)elS+ is 6 X 1Ogs),16and the nitrogen

0022-2623/93/1836-1839$04.00/0 0 1993 American Chemical Society

Ware et al.

1840 Journal of Medicinal Chemistry, 1993, Vol. 36, No. 13

mustard ligand would be displaced only very slowly. Since the Co(III)-Co(II)reduction potential can fall in the range of cellular reductants (-200 to-400 mV us NHE), chemical or metabolic one-electron reduction of the inert Co(II1) complexes would be expected. The resulting labile Co(I1) species (4) would undergo very facile ligand substitution by water,17releasing the cytotoxic free nitrogen mustard and [Con(H&)6l2+(5) (Scheme Ib). To ensure hypoxic selectivity, the reduced Co(I1) complex containing the nitrogen mustard ligand must be sufficientlystable to allow reoxidation18Jgin oxygenated cells to compete effectively with ligand loss. The overall process is essentially irreversible, since [CoI1(H2O)sI2+ is highly stable with respect to reoxidation (Eo = +la00 mV). This chemistry is analogousto that describingthe oxygen-sensitivereduction of nitro aromatic compounds (Scheme Ia) in that in oxygenated cellsreoxidation of the [ConLd2+intermediate provides a "futile" biochemicalcycle capableof suppressing net reduction. A number of Co(II1) complexes of the monodentate monoalkylator aziridine are and we have carried out a preliminary evaluation of some of these (e.g., 6).21 A Co(II1)complex (7) of the monodentate bisalkylator bis(2-chloroethy1)aminehas recently been prepared and shown to be a radiosensitiser of hypoxic ~ells.~s However, these complexes with monodentate alkylating ligands appear to lack selectivity for hypoxic cells as cytotoxins, probably because they provide Co(I1) complexes of insufficient stability for reoxidation by free oxygen to compete with ligand release. Since the kinetic stability of Co(I1)complexesis greatly increased if chelatingligands are used, we report in this paper the synthesis and characterization of a number of Co(II1) complexes of bidentate bisalkylating nitrogen mustards and the evaluation of these complexesas HSCs. A preliminary account of some of this work has been published.24

7

-

*'

CI

9 (DEE)

8 : X CH, (BEE) 10 : X I CHzCl (BCE)

Chemistry NJV-Bis(2-chloroethy1)ethylenediamine(DCE; 11) was prepared%by treatment of N-acetylethylenediamine (23) with excess oxirane to give the diol 24.26 This was deacetylated with concentrated HC1and converted to the mustard with SOC12 (Scheme11). NJV-Bis(2-chloroethyl)ethylenediaminedihydrochloride(BCE;10) was prepared similarly27by treatment of commercially-availableN P bis(2-hydroxyethy1)ethylenediamine with SOC12. Most Co(II1)complexes undergo very slow substitution at the inert metal center, rendering the synthesis of complexescontaining very reactive ligands difficult. The

Scheme 11. OH

CI

11 :WE

2)

a (i) oxirane/H20/5 OC/24 h; (ii) concentrated HCVSO OC/20 h; (iii) SOC12/20 OC/24 h.

Scheme 111. 3-

26:R.H

27:R=Me 28:R-Et 29 : R Pr

-

1

+

1922 CI

(i) Na(Racac)/H20/5 OC/12 h; (ii) SCE.2HCl/NaOH/H2O/charOC/20 coal/20 OC/20 min; (iii) BCE.2HCl/NaOH/H2O/charcoal/20 min.

preparation of the Co(II1) complexes of the bidentate mustards required a cobalt-ligand systemwhich undergoes relatively rapid substitution at Co(III), since the nitrogen mustards are unstable when in the required free base form. Complexes containing the isosteric but nonalkylating diaminesNfl-bis(2-ethy1)ethylenediamine(BEE; 8) and NJV-bis(2-ethy1)ethylenediamine (DEE 9) were also prepared, both as noncytotoxic model compounds and for use as analogues of the more reactive mustards in the developmentof suitable syntheticroutes. Suitable cobalt(111) precursor complexes are the series tram-Na[Co(Racac)2(NO2)2](26-29) (R = H,%Me, Et, n-Pr; Racac = 3-alkylpentane-2,4dionatoanion), each prepared by treatment of Na&o(NOz)6] with Na[Racacl. The reaction of the free bases of the chelating ligands with the Na[Co(Racac)2(NO2)2lcomplexes in the presence of activated charcoal (Scheme 111)gave good yields of the complexes of the diamines with the symmetrical mustard BCE, but the unsymmetrical mustard DCE was less stable in the free base form and gave considerably lower yields of the complexes (Table I). The Clacac complex (18) wae prepared from the corresponding acac complex (14) by direct chlorination with N-chlorosuccinimide.29 In most cases the complexes could be purified by direct crystallization of the perchlorate salts, though cation-exchange chromatography was necessary for the DCE complexes, 21and 22, and for the Clacaccomplex, 18. When necessary, the more solublechloridesalts could be generatedby anionexchange chromatography. In some instances, final purification was effected by reverse-phase HPLC. The structures of the Co(II1)complexeswere established by combustion elemental analysis and UV/vis, IR, and

Hypoxia-Selective Antitumor Agents

Journal of Medicinal Chemistry, 1993, Vol. 36, No. 13 1841

Table I. Physicochemical and Biological Data of Alkylating Ligands and Their Corresponding Co(II1) Complexes

+

+,

I

B no. struct X 8 9 10 11 12 A 13 B 14 C H 15 C Me 16 C Et 17 C Pr 18 C C1 19 D H D Me 20 21 D Et 22 D Pr 30

formula

analyses

Cl&ImN204C0C104 C,H,N,Cl C,H,N ClaHsoNZ04C~C104 C ~ & I & ~ ~ N ~ O ~ C O - C ~C,H,N,Cl O~ C,H,N Cl&&l2N204CoC104 C&&zNzOiC0-C104 C,H,N,Cl CaH&l2N204Co4101 C,H,N C & & ~ N ~ O ~ C O C ~ OC~, H , N C I & & ~ ~ N ~ O ~ C O - C ~CO, H ~ ,N C&zClzN20~CoC104 C, H, N, C1 C & & ~ ~ N # ~ C O C ~ . ~ H ~C, O H, N, C1 CaH&lzNaOC@ClO4 C,H,N RB 61458

c?

c

D

Elpa ICs0 (air)* AA8 HF(air)c AA8/UV4 CTlod(UV4) air/N2 ratioe (UV4) 2400 f 600 0.7 f 0.2 19OOO f 3000 0.99 f 0.07 30 f 4 29f2 1.5 f 0.3 53 f 7 0.18 2.0 -510 >5000'g NDf 0.9 f 0.2 -410 34fO.8 -310 890f 160 14 f 4 21500 1.6 2.7 f 0.4 >2720 >1 -42 6 5 0 f 1 7 0 -460 139f18 4.8 f 1.4 -500 18.5f3.3 3.5 f 0.7 -135 2 6 f 2 . 0 13 f 4 -235 3.1f0.8 64 f 15 0.64 f 0.17 1.9 f 0.2 -305 4.6 f 0.6 48 f 8 2.4 f 0.6 20 f 4 -350 2.3 f 0.4 20 f 3 0.51 f 0.15 4.9 f 1.9 -385 1.35f0.06 31f9 145 5.9 6600 504

*

Ella = peak potential us NHE for the reduction, from square-wavevoltammetry. ICW= concentration of drug (pM) to inhibit the growth of AA8 cella in culture to 50% of controls, using the protocol detailed in refs 35 and 41; values are means of at least 3 determinationsf SEM. under aerobic conditions. d Concentration (pM) X time (h) for 90% kill in HF = hypersensitivity factor = the ratio IC~O(AA~)/ICW(UV~) aerobic W 4 stirred cell suspensions. e Aerobic CTlo/hypoxic CTlo. f Nontoxic at the solubility limit. g RB 6145 employed as positive control. h For a 1-h exposure.

NMR spectroscopy. lH and l3C NMR was an important tool for distinguishing the three diastereomers formed when the symmetricalBCE is coordinatedto Co(II1).Upon coordination each nitrogen atom becomeschiral,producing RR, SS,and meso-RS isomers. In general the more stable form (RS)was enriched by fractional crystallization, and NMR analysis was used to confirm the isomeric purity. An X-ray structure determination on [Co(Clacac)a(BCE)IClOr (18) confirmed the expected structure." The stereochemistry at each chiral nitrogen atom confirmed the stereochemicalassignmentsmade by NMR. Assignments of 1Hand13C NMRspectra were establishedfrom chemical shift considerations together with 2-dimensional ('H-lH and lH-13C) NMR experiments on selected compounds. Results and Discussion Physicochemical Properties. The bidentate Co(II1) complexesprepared in this work are listed in Table I. The electrochemical properties of these complexes were investigated by cyclic and square-wave voltammetry. As observed previously21for Co(II1) aziridine complexes, reduction to Co(II) in aqueous solution is generally an irreversible process that is ascribed to rapid ligand loss from the labile Co(I1) center.30 As a result, the Co(III)/ Co(I1) reduction as measured by cyclic voltammetry is usually electrochemically irreversible. However, it was found that electrochemicalmeasurementsof the reduction processes of the bidentate mustard complexes in CHzClz (0.15 M tetrabutylammonium perchlorate) at a platinum electrode showed quasi-reversible behavior. This is consistent with slower ligand substitution at Co(II), relating to both the presence of bidentate ligands on cobalt and the use of the nonnucleophilic solvent, CH2C12. Similar behavior has been reported for other tris(8-diketonato)cobalt(III)/(II) couples.31 Reduction potentials for the nitrogen mustard complexes were determined by squarewave voltammetry, using the peak potential of the reduction wave referenced to the internal ferrocinium/ ferrocene couple.

The Co(III)/Co(II)reductionpotentials of the complexes (12, 13) with the nonalkylating ligands showed the

significant effect of varying the structure of the amine ligand, with the complex (12) of the symmetrical diamine BEE having a much lower potential (-510 mV versus the normal hydrogen electrode;NHE) than that of the complex (13) of the asymmetrical diamine DEE (-410 mV). This structural difference was reflected in the reduction potentials of the analogous alkylating complexes, with [Co(acac)z(BCE)]+ (14) having a potential (-310 mV) approximately 125mV lower than that of [Co(acac)z(DCE)]+ (19). A useful way to fine tune the reduction potentials within each series was by variation of the substituent R in the 3-position of the auxiliary Racac ligands. More electron-donating groups (R = C1< H < Me < Et < Pr) increase the effective charge on the oxygen atoms of the Racac ligand coordinated to Co(III), which in turn stabilizes the higher oxidation state with respect to reduction to Co(I1). This can be seen in the trend in Ell2 values in the series (14-18) (Table I). Although the lipophilicities of the complexes were not measured, variation of the 3-substituent also allows significant variation of this parameter. We have previously shown that coordination of reactive nitrogen species such as aziridine to Co(II1) stabilizes the ligand, even in the presence of strong acid. The bidentate Co(II1) complexes showed similar stability in neutral solution, showing no detectable change (as monitored by 1H NMR) over periods of weeks at room temperature. Biological Activity. The cytotoxicity of the ligands and their corresponding complexes were evaluated in a growth-inhibitionassay against both the CHO-derivedcell line AAV2 and the subline UV4 (Table I). The latter is a DNA repair mutant with a marked hypersensitivity to alkylating agents which cause cell killing by bulky DNA and the ratio of monoadducts or interstrand cros~-links,3~ ICs0 values in these two lines (the hypersensitivity factor thus provides valuable inHF = IC~O(AA~)/ICW(UV~)) formation about the mechanism of cytotoxicity.M*36The

Ware et al.

1842 Journal of Medicinal Chemistry, 1993, Vol. 36, No.13 1oo

lo-'

h

2a,

.r(

0 ..-I rw

*

w

TO-' ' -

0

1

2

3

0

2

4

6

8

0

Cmpd 20

10-3/

0

1

2

3

I

1

I

2

3

Cmpd 21

1 1

2

3

0

1

2

3

0

1

Time ( h r ) Figure 1. Cytotoxicity of selected compoundsagainst UV4 cells under aerobic (open symbols)and hypoxic (fiedsymbols) conditione in stirred, continuously-gassedcell suspensions. Symbols: ( 0 , O ) controla; 14 (0,m) 2000 p M ; (A,A) 10 OOO p M 1s (0,m) 340 p M 11 ( 0 ,m) 0.1 pM, (A,A) 0.15 p M 19 ( 0 )1.25 p M , (A)1.8 p M , (W) 0.3 pM, (A)0.5 pM,(V)0.8 p M 20 ( 0 )1 pM,(A)1.6 pM,(v)2

pM,(m) 0.15 pM,(A)0.2 pM,(v) 0.25 pM;21 ( 0 ) 1 pM,(A)1.8 pM,(E) 0.2 pM,(A)0.3 p M , (v)0.4 pM. two nonalkylating diamines (8,9) were relatively nontoxic be due to lower stability as the alkyl derivatives (15-17) as expected and had HF values of essentially unity, have lower reduction potentials and may be related instead indicating that their cytotoxic effects are probably not to higher lipophilicity. The Clacac complex (18) was as due to DNA adduct formation. The cobalt complexes of cytotoxic as the free ligand. Since it is likely that the these nonalkylating ligands showed very different aerobic mustard will be equally deactivated in this complex, the higher aerobic toxicity may be due to its more ready cytotoxicities inAA8 cell cultures. The BEE complex (12) was relatively nontoxic as expected, but the high toxicity reduction, as signaled by its high reduction potential (-135 of 13 was a surprise. The latter compound also showed mV). The DCE complexes (19-22) proved much more an HF of about unity, indicating no direct DNA-alkylating cytotoxic in aerobic AA8 cultures, with ICw values little activity. different from that of the free ligand. This is likely due The two alkylating ligands BCE (10) and DCE (11) were to the more positive reduction potentials of this series orders of magnitude more cytotoxicthan the corresponding (Eli2 for the DCE complex (19) is -235 mV, compared nonalkylating amines and had high HF values indicative with -310 mV for the BCE complex (14)). of DNA cross-linking activity, as expected. The nonsymThe hypoxic selectivities of selected compounds (those metrical DCE (11) was 15-fold more potent in the ICs0 with reduction potentials in the appropriate range, apassay; the most likely reason for this is higher reactivity proximately -300 to -460 mV) were determined in UV4 rather than geometrical factors. In tissue culture medium cell cultures, using the stirred suspension culture assay at 37 "C, 11 had a half-life of ca. 15 min (measured by (Table I and Figure 1). The concentration X time to reduce bioassay),%whereas 10 was much more stable (little loss cell survival to 10% (CTlo) was used as a measure of after 2 h). The biological activities of the two series of cytotoxic potency under aerobic and hypoxic conditions (Table I). cobalt complexes of the alkylatingligands (BCE and DCE) provide considerable insight into the mechanism of action The parent BCE complex (14) showed very low potency and required properties of this type of compound. In both in this assay (CTlo = 21 500 pM-h), as it did in the IC60 series,the HF values of the cobalt complexes were broadly assay. However, it showed some preferential toxicity in similar to those of the respective free ligands, those of the hypoxic cultures, with the ratio of CTlo values (air/Na) DCE complexes (19-22) being much larger (20-64) than being 1.6 (Figure 1). The methyl analogue (15) was those of the BCE complexes (14-18) (3-14). This suggests selectively toxic to hypoxic cells but was not sufficiently soluble to determine aerobic cytotoxicity so the extent of that the cytotoxicity of both series of compounds is due to release of the free ligands. its selectivity could not be determined. In contrast, the The cobalt-BCE complex (14) was much less toxic than DCE complexes (19-22) were much more potent (CTlo the parent ligand toward aerobic AA8 cells (IC60 = 890 values ca. 1pM-h), although in each case potency was less than for the free DCE ligand. DCE itself showed weak versus 30 pM), suggesting both that complexation does hypoxic selectivity, as has been reported for other nitrogen inactivate the mustard and that 14 is reasonably stable under aerobic conditions in cell culture despite quite a mustards.gB*37The parent DCE complex (19) showed high reduction potential. Other members of the series hypoxic selectivityno greater than that of the free ligand, have higher aerobic cytotoxicities, which are not likely to but the methyl analogue (20) showedsubstantial selectivity

Hypoxia-Selective Antitumor Agents

Journal of Medicinal Chemistry, 1993, Vol. 36, No. 13 1843

((20 f 4)-fold over three independent experiments),B and the ethyl analogue (21) was also significantly selective, although with a lower ratio (Table I and Figure 1). The results may indicate a narrow range of acceptable reduction potential, with an optimum close to that for the methyl analogue (20) ( E l p = -305 mV), although the relationship between the (nonthermodynamic) Ell2 values and rates of reduction (or ita reversibility by oxygen) has not yet been established for compounds of this type. The decrease in rates of cell killing with time seen in Figure 1suggests that all three DCE complexes investigated are metabolically labile under hypoxic (but not oxic) conditions, with no obvious dependence of this on E l p . No such lability is evident for the BCE complexes (14, 151, although the stability of the released BCE ligand would preclude detection of metabolic reduction of these complexes by this method. On the basis of this limited range of examples, the DCE complexes appear to have hypoxic selectivity superior to that of the BCE compounds. The reasons for this are not clear but may relate to the much higher cytotoxicity of the DCE ligand, which may ensure that ligand release rather than redox cycling is the dominant mode of cytotoxicity. Studies in progress indicate that the hypoxic selectivity of 20 is not restricted to the repair-deficient UV4 cell line, since i t shows similar selectivity for AA8 and EMTG cells under hypoxic conditions. It also has high activity against intact EMTG spheroids, suggesting that the released DCE is capable of back-diffusion from the hypoxic core of the spheroid.39 This compound is currently under investigation in uiuo as the lead compound of this new class. Conclusions The cobalt complexes of nitrogen mustards offer an attractive alternative chemistry to nitro aromatic compounds as a design for HSCs, since obligate one-electron reduction to the Co(II) species results in the rapid release of very reactive aliphatic mustards. The critical point was whether the reduced Co(1I)Lecomplex could be made sufficiently stable to allow reoxidation in aerobic cells to compete with ligand release. Previous work22*29 suggested that monodentate alkylating nitrogen ligands could not provide sufficient stability. The results shown here provide valuable insight into this and other questions relating to the design of metal complexes as HSCs and suggest that chelating diamines can provide complexes of sufficient stability to allow reoxidation of the labile Co(I1) species to compete with ligand release. The pattern of hypoxic selectivities shown by the DCE seriesof complexes (19-22) suggests the critical importance of the redox potential, which presumably controls the rate of bioreduction, and implies a quite narrow optimal range. This work demonstrates that metal complexes of nitrogen mustards have significant hypoxia-selective cytotoxicity toward mammalian cells in cell culture. The compounds have the capability of releasing mustard ligands which are reactive enough to be highly cytotoxic and yet possess a long enough half-life to diffuse from the hypoxic cells where they are activated to surrounding cells at higher levels of oxygen tension.' This work is a f i s t

performed on SP Sephadex (2-25 (Pharmacia) in the Na+ form. HPLC purifcations were performed using a Waters 800 quaternary pump withWISP injector and Gilson 202 fraction collector controlled by an HP chemstation and using an HP 1040Adiodearray spectrometer as detector. lH and 'BC NMR spectra were recorded at 400 and 100 MHz, respectively, on a Bruker AM 400 spectrometer. Chemical shifts are reported relative to internal Me&. High-resolution mase spectra (FAB+)were recorded from a 3-nitrobenzyl alcohol matrix on a VG 70-SE maas spectrometer using argon gas. Electrochemical measurementswere performed onaBioanalyticalSystemsBAS 100Ausingthesoftwarepackages provided for cyclic voltammetry and Osteryoung square-wave voltammetry. Tetra-n-butylammoniumperchlorate electrolyte was twice recrystallized from EtOAc and dried in vacuo at 80 "C. Melting points are uncorrected. Elemental analyses were performed by the Microchemical Laboratory at the University of Otago, Dunedin, New Zealand. Nfl-Bis (2-chloroethy1)ethylenediamine Dihydrochloride (BCE2HC1,lO). A solution of NJP-bis(2-hydroxyethy1)ethylenediamine(5.40g, 0.036 mol) in SOCl2 (60 mL) was heated at 90 "C for 1 h and then left at room temperature for 24 h. Excess SOClz was removed under reduced pressure, and the residue was triturated with 2-propanol. Recrystallization from boiling 2-propanol (800mL) to which enough Hz0 (ca. 5 mL) had been added to effect complete dissolution of the solid gave N," bie(2-chloroethyl)ethylen~edihydrochloride (5.95 g, 63% ) (BCE.2HC1; 10) as plates. lH N M R (DsO): 6 3.93 (m, 4 H, CHr NMR Cl), 3.58 (8, 4 H, CHzNHR), 3.58 (m, 4 H, CHZCHzCl). (DaO): 6 51.96 (CHZCH~C~), 45.56 (CHzNHR),41.74 (CHpC1). N ~ - B i s ( 2 - c h l o r o e t h y l ~ tDihydrochloride ~~n~e (DCE2HC1,ll). Oxirane (27.0g, 0.60 mol) was addedto a cooled (5"C) solution of N-acetylethylenediamine(23) (25.0 g, 0.24 mol) in water (50 mL). The solution was stirred for 4 h at 6 O C and then overnight at room temperature before being concentrated under reduced pressure. The residue was chromatographed on SiOz, and elution with EtOAc/MeOH (91) gave N-acetyl-N'JV'bis(2-hydroxyethy1)ethylenediamine(24) am a viscous oil (35.7 g, 78%). This was used directly, the entire sample beimg dissolved in concentrated HCl(250 mL), warmed to 90 "C for 20 h, and then concentrated under reduced pressure to give the dihydrochloride salt of N~-bis(2-hydroxyethyl)ethylenediamine(25) as a syrup which slowly crystallized, mp 116-120 "C (&.amp 116-118 O C ) . This was dissolved in MeOH, and the solution was neutralized with powdered KHCOs, fiitered,and evaporated. The reaiduewastrituratedwithMezCO/MeOH(l:l),andthetriturate was evaporated to give the corresponding free base as a strawcolored liquid, which was used without further chmacterization.a A solution of this (2.96 g, 0.02 mol) in SOClz (150mL) was stirred at room temperature for 48 h. Excess SOClz was then removed under reduced pressure, and the residue was dissolved in water and washed several times with EtOAc. The aqueous layer was evaporatedto drynew under reduced pressure, and the resulting crude residue was crystallized from MeOH to give NJV-bis(2chloroethy1)ethylenediamine dihydrochloride (2.64 g, 49 % ) (DCE.2HC1; 11) as hygroscopic white plates, mp 136 "C (lit.smp 139-140 OC). 'H N M R (DzO): 8 4.03 (t,J = 5.7 Hz, 4 H, CHZCl), 3.74 (t, J 7.7 Hz, 2 H, 3.81 (t, J = 5.7 Hz, 4 H, CH~CHZCI), C H a W ) , 3.55 (t,J = 7.7 Hz, 2 H, CHzNHI). "C NMR (DaO): 6 57.55 (CHzCHzCl), 52.54 (CHaW), 39.82 (CHzCl). Na[Co(Meacac)s(NOa)aI.O.SHaO(27). NadCo(NOz)~1(3.27 g, 8.11 mmol) was dissolved in HzO (11 mL) and added to a mixture of NaOH (0.70 g, 17.5 mmol) and 3-methyl-2,4-pentanedione (2.0 g, 17.5 mmol) in Ha0 (11mL) which had been cooled in an ice bath. Rapid formation of red-brown crystals occurred after 10 min, and after cooling at 5 O C for 12 h these were collected by fiitration and washed with MeCO and EO0 and dried in air to give 27 (2.81 g, 82.9 % 1. Thiswas recrystallized by dissolving 1g in HzO (35mL) and fitering intoNaNOz solution (5 g in 15 mL of BO). The resulting crystallime product was washed with EtOH/Me&O (2:l) and dried in air to give Na[Co(Meacac)~(NO~)d.0.5H~O (27) (0.40 g, 40%). Anal. (CuHleN~OsNaCo-0.5H20)C, H, N. SodiumBis(3-ethyl-2,4-pentonedionato)dinitroco (111) Hydrate Na[Co(Etacac)~(NO1)a].H10 (28). Prepared as described for 27 using Na&2o(NOz)el (3.28 g, 8.12 mmol),

step toward defining the parameter limits (particularly metal redox potential and released mustard reactivity) for this new general class of hypoxia-selective cytotoxins.

Experimental Section Reagents and Physical Mearurements. NJP-Bie(2-hydroxyethy1)ethylenediamineand N-acetylethylenediaminewere obtained from Aldrich. Cation-exchange chromatography was

1844 Journal of Medicinal Chemistry, 1993, Vol. 36, No. 13

Ware et al.

3-ethyl-2,4-pentanedione(2.37 g, 18.5 mmol), and NaOH (0.74 3.08 (dd, J = 7.8,3.7 Hz, 2 H, CHzNHR), 2.76 (dd, J = 8.3,ll.O g, 18.5 mmol) to give Na[Co(Etacac)~(N0~)~1~H~0 (28) (2.91 g, Hz,2 H, CHZNHR),2.39,2.16 (m, 2 H, CHzCHzCl),2.10,2.16 (8, 3 CH3CO),2.26 (t,J = 7.8 Hz, 4 H, CHZCHZCHS), 1.38 (m, 4 H, 80.3%). 0.95 (t,J = 7.3 Hz, 6 H, CHs). "C NMR (CDCh): 6 Sodium Bi~(3-n-propyl-2,4-pentanedionato)dinitrocobal- CH~CHS), 189.39, 189.24 (CO), 109.00 (CPr), 50.89 (CHZCl), 49.80 (CHZtate(II1) Hydrate Na[ Co(Pracac),(NOa)r].HrO (29). PreNHR),39.67 (CHzCHZCl),31.19 (CH&H&H3), 25.80,25.28(CHspared as described for 27 by mixing Na[Co(NO2)61 (1.0 g, 2.48 CO), 24.21 (CHzCHa), 13.86 (CHs). Anal. ( C ~ ~ C 1 2 N & mmol) in HzO (2.5 mL) and Na[PracacI.HzO (0.94 g, 5.16 mmol) Cd104) C, H, N. in HzO (5 mL). After 5 min the solution was fiitered to remove Bis(3-chloro-2,4-pentanedionato) (N,N-bis(2-chloroetha brown insoluble impurity, MeOH (2 mL) was added, and the yl)ethylenediamine)cobalt(III) Chloride [Co(Clacac),(BCE)]solution was left at room bmperature for 48 h to give Na[CoCl(18). N-Chlorosuccinimide(0.204 g, 1.53mmol) was dissolved (Pracac)z(N02)~I.HzO(29) (0.23 g, 19.6%1. Anal. (CleHZsN20~NaCoeHzO) C, H, N. in MeOH (60 mL). [Co(acac)~(BCE)lC104(14) (0.24 g, 0.443 mmol) was added portionwise, and the solution was stirred for Bis(2,4-pentanedionato)(N,N-bie(2-chloroethyl)ethyl6 h a t 20 OC. The solvent volumewas reduced to 30 mL, and HzO enediamine)cobalt(111) Perchlorate [Co(acac)t(BCE)]ClO~ (50 mL) was added. The solution was loaded on to a Sephadex(14). Na[Co(acac)z(NO2)z].HzO (26)2a(0.40 g, 1.025 mmol) was SP-C25 cation-exchangecolumn (2.5 X 10 cm) prepared in the dissolved with stirring in a mixture of HzO (6 mL) and MeOH Na+form. The column was washed with water, and the complex (2 mL). An ice-cooled solution of BCE.BHCl(10) (0.279 g, 1.081 was eluted with 0.1 N NaCl. The eluted band was extracted five mmol) dissolved in Ha0 (2 mL) was neutralized by the addition times with CHZC12, and the combined extracts were evaporated. of 2.0 mL of a solution of NaOH (0.43 g) in MeOH (10 mL). Toluene (5 mL) was added to the residue, and the solution was Immediately,activated charcoal (0.25 g) was added to the stirred further evaporated to give a magenta-colored oil. Addition of solution of cobalt complex (26),followed rapidly by the solution MezCO (3 mL) produced a mass of fine needles of [Co(Clacac)zof deprotonated BCE. The mixture was stirred for 20 min and (BCE)IC1(18)(0.15 g, 55.5%),which were fiitered and washed then fiitered through Celite, and the charcoal was washed once quickly with MezCO followed by Et20 and dried in air in a with water and once with MeOH. The washings were added to desiccator. 'H NMR (CDClS): 6 5.86 (br, 2 H, NH), 3.97 (m, 4 the filtrate, followed by NaC104.HzO (3.2 g) in HzO (3 mL), and H, CHZCl), 3.00 (m, 4 H, CHzNHR), 2.95 (m, 4 H, CHZCHzCl), the mixture was cooled in an ice bath. After 2 h the purple 2.49, 2.42 (8,3 H, CHsCO). "C NMR (CDCh): 6 188.74,188.45 crystalline mass was filtered and washed twice with cold H20 (CO),107.06 (CCl),51.10 (CHzCl),50.16 (CHzNHR),39.92 (CHzand three times with Et20 and dried in air to give [Co(acac)zCHzCl), 28.26,27.60 (CHsCO). Anal. (C&&~NZO~CO-C~) C, (BCE)]ClO4 (14) (0.475 g, 81%). 'H NMR ((CD3)zSO): 6 5.97 H, N. (br, 2 H, NH), 5.66 (s,2 H, CH), 4.03,3.93 (m, 2 H, CHzCl), 2.86, 2.75 (m,2H,CH~NHR),2.71,2.58(m,2H,CH2CH~C1),2.14,2.08 Bisl2,4-pentanedionato)( N ~ - d i e t h y l 9 t h y l e n ~ ~ i n e ) (a, 3 H, CH3CO). l3C NMR ((CDs)ZSO): 6 189.47, 189.31 (CO), cobalt(B1) Perchlorate [Co(acac)t(BEE)]ClO, (12). This 97.93 (CH), 50.72 (CHzCl), 49.70, (CHzNHR),39.60 (CHzCHzcomplex was prepared as described for 14 using Na[Co(acac)r Cl), 26.38,26.25 (CHsCO). Anal. (ClsHzaClzNz04Co.C104) C, H, (NOa)&H20 (26) (0.70 g, 1.794 mmol) and BEE (0.25 g, 2.151 N, C1. mmol) to give [Co(acac)z(BEE)IClO4(12) (0.75 g, 88.4%). 'H Bie(3-methyl-2,4-pentanedionato)(N~-bis(2-chloroeth- NMR (CDCls): 6 5.51 (s,2 H, CH), 4.57 (br, 2 H, NH),3.07 (br q,J = 3.8 Hz, 2 H, CHZNHR),2.62 (m, 2 H, CHZNHR),2.40,1.88 y1)ethylenediamine)cobalt (111)Perchlorate [Co(Meacac),(m, 2 H, CHZCHS), 2.20,2.09 (s,3 H, CHSCO),1.32 (t,J = 7.3 Hz, (27) (BCE)]ClO4(15). Reactionof Na[Co(Meacac)~(NOz)21-.H~O 6 H, CEfaCHz). '3C NMR (CDCL): 6 190.55,190.37 (CO), 98.63 (0.50 g, 1.196 mmol) with BCE.2HC1 (10) (0.34 g, 1.318 mmol) (CH),49.97 (CHzNHR),44.63(CHzCHs), 26.36,26.27 (CHsCO), as described for 14 gave [Co(Meacac)z(BCE)]C104 (15) (0.40 g, 12.22 (CH3CH2). Anal. (C&&O~CO*C~O~)C, H, N, C1. 58.7%). 13CNMR(CDC4)(majorisomer): 6 188.65,188.20 (CO), 102.25 (CMe), 50.70 (CHzCl), 49.70 (CHzNHR),40.07 (CH2CHr Bie( 2,4-pentanedionato)(N~-diethylethylenedine)Cl), 26.65, 26.33 (CHsCO), and 14.95 (CH3). Anal. (C&p cobalt(II1) Perchlorate [Co(acac),(DEE)]C104 (13). This ClzNzO&o*C104)C, H, N. complex was prepared as described for 14 using Na[Co(acac)r Bis(3-ethyl-2,4-pentanedionato)(N,N-bi~(2chloroe~thyl)- (NO2)2]-HzO(26) (0.70 g, 1.794 mmol) and DEE (0.24 g, 2.065 (13) (0.52 g, 61.3%). 'H mmol) to give [Co(acac)~(DEE)IC104 ethylenediamine)cobalt(III) Perchlorate [Co(Etacac)aNMR (CDCl3): 6 5.60, 5.55 (8, 1 H, CH), 4.36, 3.99 (br 8, 1 H, (BCE)]ClO4 (16). This was prepared as described for 14 using Na[Co(Etacac)z(NOz)a.HZO(28)(0.80g, 1.868mmol),BCE~2HC1 NHz), 3.22, 2.50, 2.23, 1.91 (9, 1H, J = 6.8 Hz, CHzCHs), 3.13 (br s,2 H,CH~NHZ), 2.74 (br d, 2 H, J = 4.7 Hz, CHzNRp), 2.22, (0.48 g, 1.868 mmol), and NaOH (0.14 g, 3.6 mmol). Crystals of (16) (0.585 g, 52.4%) were isolated by 2.18, 2.11, 2.02 (8, 3 H, CHsCO), 1.17,0.95 (t, 3 H, J 6.8 Hz, [Co(Etacac)~(BCE)IC104 slow evaporation of the MeOH/H20 solution. 1HNMR (CDCl3): CH3CH2). "C NMR (CDCh): 6 190.82, 190.75, 190.29, 189.78 (CO),99.56,98.32(CH),60.87 (CH2NR2),47.92,26.61(CHzC&), 64.88 (br s, 2 H, NH), 3.89, 3.78 (m, 2 H, CHZCl), 3.08 (br q, 2 H, CHZNHR),2.76 (t,2 H, CHZNHR),2.40, 2.19 (m, 2 H, CHZ41.71 (CHzNHz), 26.52, 26.49, 26.41, 25.78 (CH&O), 8.64, 7.58 (CHsCHz). Anal. ( C ~ ~ H & J ~ O ~ C OC,CH, ~ ON.~ ) CHzCl), 2.35, 2.17 (8, 3 H, CHsCO), 2.32 (a, 4 H, J = 7.4 Hz, CH~CHB), 1.02 (t,6 H, J = 7.4 Hz,CHsCHz). '9c NMR (CDCls): Bie(2,kpentaneclionato)(N~-bis(2-~hlo~hyl)ethylen~ 6 189.26, 189.13 (CO), 110.49 (CEt),50.91 (CHzCl),49.82 (CHzdiamine)cobalt(III)Perchlorate [Co(acac)t(DCE)IC1O4(19). NHR), 39.69 (CH&H2Cl), 25.56, 25.00 (CHsCO), 22.37 (CH2This complex was prepared as describedfor 14,using a SephadexCH3), 15.17 (CH3CHz). Anal. (Cd~Cl~NzO&o.C10r) C, H, N, SP-C25 column to purify the product. Alternatively, the free c1. base of DCE was prepared by suspending DCE.BHCl(O.364 g, Bis(3-n-propyl-2,4-pentanedionato)(N~-bis(2-chloroet- 1.41 mmol) in MeOH (5 mL) and adding NaOH (0.113 g, 2.82 hyl)ethylenediamine)cobalt(III) Perchlorate[Co(Pracac)ammol) in MeOH (2 mL). The resulting solutionwas immediately (BCE)]C104 (17). This complex was prepared as described for added to a solution of Co(acac)s (0.457 g, 1.28 mmol) in MeOH 14 using Na[Co(Pracac)2(NOz)&HzO(29) (0.500g, 1.054 mmol), (45 mL), followed by activated charcoal (0.1 8). The solution was stirred for 1hand then filtered through Celite. The combined BCE.2HCl(O.272g,1.054mmol),andNaOH (O.O84g, 2.108mmol). green-red filtrate and washingswere evaporated to small volume After the reaction mixture was filtered to remove the charcoal, under reduced pressure, and HzO (50 mL) was added. Green the solution was acidified with 0.1 N HCI and extracted with crystals of weacted Co(acac)swhich formed were fiitered off, CHCL. The CHCla solution was evaporated to dryness, and the residue was dissolved in MeOH/HZO (1:3), loaded on to a and the filtrate was then loaded on to a Sephadex-SP-C25column Sephadex-SP-C25column in the Na+ form, washed with HzO, (60 mL) in the Na+ form. The column was washed with water, and elution was performed with a gradient of 0.05-0.1 N NaC1. and eluted with aqueousNaCl solution (0.15mol L-'). The eluant containing the product was extracted with CHCl3, and the The eluted band was extracted with CHCh four times, and the resulting solution was evaporatedto dryness,giving [Co(Pracac)zcombined extracts were evaporated to dryness under reduced (BCE)]Cl(O.365g, 60.1 %). This was converted to the clod- salt pressure. The residue was taken up in water, and NaC104eH20 (1g) in MeOH was added. After cooling at 5 OC for 2 days, the (95% yield) by addition of NaClO4 to a MeOH/H20 solution of dark crystals of [Co(aca~)~(DCE)lC104 (19) which had formed the complex, followed by slow evaporation to give dark greenpurple dichroic needles of [Co(Pracac)~(BCE)lClO4(17). lH (0.03 g, 4.3%) were filtered and washed with HzO and dried in NMR (CDClS): 6 4.87 (br s,2 H, NH), 3.88,3.77 (m, 2 H, CHZCl), air in a desiccator. 1H NMR (CDCL,): 6 5.60.5.53 (e, 1H, CH),

Hypoxia-Selective Antitumor Agents

Journal of Medicinal Chemistry, 1993, Vol. 36, No. 13 1845

4.39,4.22 (br m, 1H, "21, 3.93,3.69,3.59,3.50 (m, 1H, CHzC1), 3.07 (m, 2 H, CHZNHZ), 2.79 (t,2 H, J = 6.3 Hz, C H Z W ) ,3.02, 2.61,2.38,2.26 (m, 1H, CHZCHzCl),2.21,2.19,2.10,1.97 (8, 3 H, CHsCO). '9c NMR (CDCb): 6 191.51, 191.46, 191.12, 189.91 (CO), 99.81,98.46 (CH), 61.40 (CHzNRa),55.71,53.77 (CHzCHr Cl),42.11 (m&Hz),37.97,35.97 (CHzCl),and 26.68,26.64,26.26, 25.71 (CHaCO). Anal. (C&&lzNz04C0C104) C, H, N. Bis(3-methyl-2,4-pentanedionato)(NJY-bis(2-chlor~thyl)ethylenediamine)cobalt( 111)Perchlorate [Co(Meacac)l(DCE)]ClO4 (20). Na[Co(Meacac)z(NOdvHzO(27)(1.5Og,3.587 mmol) was dissolved in a mixture of MeOH (46 mL) and HzO (27 mL). A solution of NaOH (0.330 g, 8.248 mmol) dissolved in MeOH (8 mL) was added to a solution of DCE-2HCI (1.064 g, 4.124 mmol), in HzO (1mL) which had been cooled in an ice bath. After 30 8, activated charcoal (0.42 g) was added to the Na[Co(Meacac)2(NOz)&HzO solution, followed immediatelyby the deprotonated DCE solution, and the mixture was stirred for 1 h. The charcoal was filtered off through Celite and washed with MeOH, which was added to the filtrate. The filtrate was acidified with 3 mol L-l HCl(l.5 mL) and extracted with three portions of CHCb. The combined extracts were evaporated to dryness under reduced pressure, and the residue was taken up in H20 (20 mL) and decanted from some insoluble material. MeOH (20 mL) was then added to the supernatant, and the solution was left open to the air at 20 "C for slow evaporation of the MeOH. After 1week, the resulting green crystals of [Co(Meacac)z(DCE)]ClO4 (20) (0.205 g, 10%) were collected by filtration and washed with 20% MeOH/HzO, HzO, and finally EhO. The product was air-dried in a desiccator. lH NMR (CDCb): 6 4.30, 4.01 (br, 1H, NHz), 4.07, 3.78, 3.60, 3.59 (m, 1 H, CHzCl), 3.10 (m, 2 H, CH&"H), 2.98,2.76 (m, 1H, CHzNRz), 2.61,2.50,2.25 (m, 1H, CHZCHzCl),2.35,2.33 (e, 3 H, CHs), 2.24, 2.09, 1.98, 1.90 (8, 3 H, CH&O). l9c NMR (CDCb): 6 189.42, 189.20, 188.87, 187.71 (CO),103.83, 102.16 (CMe), 61.15 (CH2NRS), 55.67, 53.60 (CHZCHaCl),41.83 (CHzNHz), 38.17, 36.03 (CHaCl), 26.46,26.38,26.24,25.64 (CHaCO), 14.89,14.69, (CHs). Anal. (Ci~szCl~N204C~ClO4) C, H, N, C1. Bis(3-ethyl-2,4-pentaneddionato) (N~-bis(2-chloroethyl)ethylenediamine)cobaIt(III) Perchlorate [Co(Etacac)t(DCE)]ClOd (21). A freshly-prepared solution of the free base of DCE (11) (0.856 g, 3.319 mmol) was added to a solution of Na[Co(Etacac)z(NOhl1.H10(28) (1.288 g, 2.886 mmol) in HzO (10 mL) and MeOH (20 mL) to which activated charcoal (0.28 g) had been added. The mixture was stirred for 1 h and then filtered through Celite, and the charcoal was washed with water and MeOH which were added to the filtrate. HC1 (3 N) waa added to the filtrate until the solution was acidic, and it was then extracted three times with CHCla. The combinedextracts were evaporated under reduced pressure to a thick oil, which was dissolved in a mixture of MeOH (10 mL) and Hz0 (10 mL) and extracted three times with CHCb (10 mL). The combined extracts were once again evaporated to an oil and then taken up in MeOH (15mL) and HzO (15 mL) and loaded on to a SephadexSP-C25 cation-exchange resin (Na+ form) column and eluted with 0.15 N NaCl in 10% MeOH/HaO. The green eluant was extracted five times with CHCL, and the combinedextracts were evaporated under reduced pressure to give an oil, which was dissolved in EhO and then evaporated to dryness to give [Corn a green solid (455mg, 29.5%). This (Etaca~)~(DCE)]C1.2H~0 was converted to the Clod- salt 21 as described for 17, extracted into CHZClz, and isolated byevaporation of the solventtodryness. 1H NMR (CDCb): 6 5.54, 4.46 (br, 1H, NHz), 4.02, 3.76 (m, 1 H, CHzCl), 3.53 (m, 2 H, CH&l), 3.43, 2.69, 2.61, 2.30 (m, 1 H, CH2CHZC1),3.17 (br 8, CHzNHz),2.93 (br m, 2 H, CHzNRz), 2.40, 2.35, 2.23, 2.09 (8, 3 H, CHsCO), 2.37, 2.29 (q,2 H, J 7.4 Hz, CHzCHa), 1.05 (t,6 H, J = 7.4 Hz, CHaCHz). 'W NMR (CDCls): 6 189.78,189.09,189.00,188.21 (CO), 111.59,109.32 (c!Et), 61.57 (CHm,55.43,53.73 (CH&HC1),41.84 (CHzNHa),37.65,36.17 (CHzCl),25.82, 25.70, 25.30, 24.55 (CH&O), 22.38, 22.32 ( C H z CHs), 15.28, 14.79 (CHsCH2). High-resolution MS (FAB): m/z 497.1359 (calcd for C&~Wl&oN~04,497.1384). Bis(3-n-propyl-2,4-pentanedionato)(N~-b~s(2-chloroethyl)ethylenediamine)cobalt(III) Perchlorate [Co(Pracac)r(DCE)]ClO4 (22). This complex was prepared as described for (29)(0.80g, 1.687 mmol), (21),usingNa[Co(Pracac)~(NOz)z]-HzO DCE-2HC1 (0.53 g, 2.054 mmol), and NaOH (0.16 g, 4.0 mmol) to give [Co(Pracac)z(DCE)]ClO, (22) (0.18 g, 17.0%) after

evaporating the solution to dryness. 1H NMR (CDCb): 6 4.21, 4.09 (br, 1H, NH,), 4.04, 3.78, 3.62, 3.57 (m, 1H, CHZCl), 3.57, 2.57,2.52,2.19 (m, 1H, CHzCHzCl),3.12 (m, 2 H, CHzNH3,3.04 (m, 1 H, CHzNRd, 2.84 (br dt, zJ = 12.2 Hz, SJ = 3.4 Hz, 1H, CHZNRZ), 2.35,2.35,2.23,2.09 (s,3 H, CHsCO), 2.30,2.22 (m, 2 H, CHzCHZCHa),1.40 (m, 4 H, CHZCHa), 0.95,0.93 (t 3 H, J 7.4 Hz, CHaCHz). "C NMR (CDCb): 6 189.97, 189.66, 189.19, 61.27 (CHaW), 55.62,53.59 188.19 (CO),110.01, 108.13 (CH), (CHzCHZCl),41.78 (CHzNHs),37.98,36.05 (CHzCl), 31.22,31.20 (CHzCHzCHs), 25.87, 25.54, 25.47, 24.72 (CHaCO), 24.27, 23.68 (CHzCHs), 14.00, 13.87 (CHsCH2). Anal. (CnH&O4ClzCec104) C, H, N. High-resolution MS (FAB): m/z 525.1677 (calcd for CaHaS6Cl&oNz04,525.1697). HPLC Purification of Cobalt Complexes. Solutions of the complexes (15, 19-21) in CH&N were chromatographed on a semipreparative C18 pBondapak column (25 X 100mm; 45pL injection), using a mobile phase of CHaCN/0.37 mol L-1 ammonium formate, pH 4.5 (52:48 v/v), at a flow rate of 3.6 mL/min. Detection was by UV absorption at 254 nm. For example,pooled eluates from 250 mg of 20, which were loaded in ca. 1.8-mg injections, were cooled and the upper, green CHaCN-rich phase was separated. The CHaCN was evaporated, and to the residual aqueous solution was added 1mol L-1 NaClO4 (2 mL) which was then extracted with CHzC12 (5 X 20 mL). After evaporation of the CHzClz and dissolution in MeOH (15 mL), 1mol L-l NaClO4 (7 mL) was added and the solution was allowed to evaporate slowly for severaldays. The green crystals (210mg) were washed with HaO, MeOWH20,and EhO. This procedure gave compound 20 of 99.7% purity based on peak area. Determination of Reduction Potentials. The Co(III)/Co(11) redox potentials were determined by Osteryoung squarewave voltammetry in CHZClz solutions approximately 10-3 mol L-l in cobalt complexand containing n-B~NCl04(0.15 mol L-1) as the electrolyte. A three-electrodeconfigurationwas used, with a Pt disk as the working electrode, a Pt wire as the auxiliary electrode,and a quasi-referenceAg/AgClelectrodeprepared from Ag wire electrolytically coated with AgC1. The ferrocenium/ ferrocenecouple was usedas internal reference (0.548 V vs "E). Cell Line Studies. AA8 and UV4 cells were maintained in logarithmic-phase growth in 25-cm2 tissue culture flasks, using antibiotic-free a-MEM with 10%v/v heat-inactivated (56 OC, 40 min) fetal calf serum. Doubling times were approximately 14 h for AA8 and 15 h for UV4 cells. Cultures were tested for mycoplasma contamination frequently, using a cytochemical staining methodsa Bulk cultures of A48 cells were prepared in spinner flash, using the above growth medium plus penicillin (100 IU/mL) and streptomycin (100 pg/mL). Growth inhibition studies were performed as describedin detail elsewhere,"*41using 200 viable AA8 or 300 viable UV4 cells plus 5OOO lethallyirradiated AA8feeder cellsper well in S w e l ltissue culture dishes. The ICs0 was determined as the drug concentration needed to reduce the cell mass (protein content, measured after 72-78 h by staining with methylene blue and measuring absorbance in a microplate photometer) to 60% of the mean value for 8 control cultures on the same 96-well plate. The stirred suspension culture assay used hae also been described in detail elsewhere." Clonogenic assays were carried out with magnetically-stirred 10-mL suspension cultures of late log phase W 4 cells at 2 X 108cells/mL. Samples were removed periodicallyduring continuous gassing with 5% COz in air or NZ at 37 "C. A range of drug concentrations were studied to identify those concentrations which gave approximately the same rate of cell killingunder both aerobicand hypoxicconditions. The ratios of the concentration X time for cell survival of 10% ( C T 3 for these two survival curves was used as the measure of hypoxic selectivity. Stability of BCE and W E Ligands. The stabilities of the nitrogen mustard ligands (LO, 11) in culture medium (a-MEM containing 5% v/v fetal calf serum) at pH 7.0 and 37 OC were investigated by bioassay againet W 4 cells,essentiallyasdeacribed previously.W Stock solutions of the mustards were prepared in 0.01N HCl on ice and diluted into mediumwhich was maintained under an atmosphere of 5% COz. Samples were withdrawn at 15-min intervals for 2 h, and suitable dilutions were titrated against logarithmic-phase W 4 cultures in 96-well plates to determine ICs0 values. The fraction of the parent compound

1846 Journal of Medicinal Chemistry, 1993, Vol. 36,No.13

Ware et al.

Jackson, T. B.; Edwards, J. 0. Coordination compounds of remaining at time t WBB determined aa the ratio I C W , ~ ~ ~ I C ~ ~ . ethylenimiine with cobalt(III), chromium(III), palladium(II), and Pseudo-fiist-order half-lives were determined by logarithmic platinum(IV). Znorg. Chem. 1962,1,398-401. regression. Ware, D. C.; Siii,B. G.; Robinson, K. G.; Denny, W. A.; Brothers,

Acknowledgment. This work was supported by the Cancer Society of New Zealand and the Health Research Council of New Zealand. The authors thank Karin Tan and Pierre van Zijl for preparative HPLC,Diane Schultz for determination of ligand half-lives, Susan Pullen for biological testing, and Wendy Hodgson for preparing the manuscript. References (1) Overgaard, J.; Hansen, H. 5.; Andersen, A. P.; Hjelm-Hansen, M.; Jorgensen, K.; Sandberg, E.; Bertelsen, A.; Hammer, R.; Petersen, M. Misonidazole combined with split c o m e radiotherapy in treatment of invasive carcinoma of larynx and pharynx: report from the D M C A 2 study. Znt. J. Radiat. Oncol. Biol. Phy8. 1989,16,1065-1068. (2) Gatenby, R. A.; Kessler, H. B.; Roeenblum, J. S.; Coia, L. R.; Moldofaky, P. J.; Hartz, W. H. Oxygen distribution in squamous cell carcinoma metastases and its relationship to outcome of radiation therapy. Znt. J.Radiat. Oncol.Biol. Phys. 1988,14,831838. (3) Olive,P. L.; Durand, R. E.; Misonidazolebmding in SCCVIItumors in relation to the tumor blood supply. Znt. J.Radiat. Oncol. Biol. Phys. 1989,16,756761. (4) Kerr, D. J.; Kaye, S. B. Aspects of cytotoxic drug penetration with particular reference to anthracyclines. Cancer Chemother.Pharmacol. 1987, 19, 1-5. (5) Moulder, J. E.; Rockwell, 5. Hypoxic fractions of solid tumors: Experimental techniques, methods of analysis, and a survey of existing data. Znt. J.Radiat. Oncol. Biol. Phys. 1984,10,696712. (6) Brown, J. M.; Koong, A. Therapeutic advantage of hypoxic cella in tumors: a theoretical study. J. Natl. Cancer Znut. 1991, 83, 178-185. (7) Wilson, W. R. Tumour hypoxk challenges for cancer chemotherapy. In Cancer Biology and Medicine. Vol. 3. The Search for New Anticancer Drugs; Waring, M. J., Ponder, B. A. J., Eds.; Kluwer: Lancaster, 1992; pp 87-131. (8) Denny, W. A.; Wilson, W. R. Bioreducible mustards: a paradigm for hypoxia-selective prodrugs of diffusible cytotoxins (HPDCs). Cancer Metastcuris Rev., in press. (9) Siegel, D.; Gibson, N. W.; Preusch, P. C.; Rose, D. Metabolism of mitomycin C by DT-diaphorase: Role in mitomycin C-induced DNA damage and cytotoxicity in human colon carcinoma cella. Cancer Res. 1990,50,7483-7489. (10) b o x , R. J.; Boland, M. P.; Friedloe, F.; Coles, B.; Southan, C.; Roberta, J. J. The nitroreductase enzyme in Walker cella that activates &(aeiridin-l-y1)-2,4dinitroImnzamide(CB 1954) to &(a&idin-l-yl)-4-hydroxylamino-2-nitrobenzamide is a form of NAD(P)Hdehydrogenase(quinone)(EC 1.6.99.2).Biochem.P h a r m o l . 1988,37,4671-4677. (11) Hajoe, A. K. D.; Winston, G. W. Dinitropyrene nitroreductase activity of purified NAD(P)H-quinone oxidoreductase: role in rat liver cytosol and induction by Aroclor-1254pretreatment. Carcinogenesis 1991,12,697-702. (12) Gustafson, D. L.; Pritsos, C. A. Bioactivation of Mitomycin C by xanthine dehydrogenase from EMT6 m o w mammary carcinoma tumors. J. Natl. Cancer Znut. 1992,84,118&1185. (13) Kutcher, W. W.; McCalla, D. R. Aerobic reduction of bnitro-2furaldehyde semicarbazone by rat liver xanthine dehydrogenase. Biochem. Pharmacol. 1984,33,79%305. (14) Maki, N.; Tanaka, N. Cobalt. Encyclopedia of Electrochemistry ofthe Elements;Bard, A. J., Ed.; Marcel Dekker: New York, 1975; pp 43-210. (15) Lewis, D. F. V. Molecular orbital calculations on tumor-inhibitory aniline mustard8 QSAR. Xenobiotica 1989,19,243-249. (16) Atwood, J. D. In Inorganic and Organometallic Reaction Mechanisms; BrookslCole: Monterey, CA, 1985; p 87. (17) Simic, M.; Lilie, J. Kinetics of ammonia detachment from reduced cobalt(II1) complexes based on conductometric pulse radiolysis. J. Am. Chem. SOC. 1974,96, 291-292. (18) Eaton, D. R.; O'Reilly,A. Oxidation of cobalt(I1)amine complexes to mononuclear cobalt(II1) complexes by dioxygen. Znorg. Chem. 1987,26,4185-4188. (19) Fallab, S.;Mitchell, P. R. Binuclear dioxygen complexes of cobalt. In Advances in Inorganic and Bioinorganic Mechanisms; Sykes, A. G., Ed.; Academic Press: London, 19W, p 326.

P. J.; Clark, G. R. Synthesis and characterization of aziridme complexes of cobalt(If1) and chromium(III) designed as hypoxiaselective cytotoxins. The X-ray crystal structure of tram-[Co(Az)r(NOa)lIBr.2HlO.LiBr.Znorg. Chem. 1991, 30,3750-3757. Ware, D. C.; Maclue, D. S.; Brothers, P. J.; Denny, W. A. Synthesis and characterization of mono- and bisaziridme bisdimethylglyoximato cobalt(III) complexes. Pol hedron, submitted. b e ,K. W.; Herman, T. S. Teicher, B. A.; Abranu, M. J.; L Cytotoxicity,radiosensitization, antitumor activity and interaction with hyperthermia of a Co(III) mustard complex. Cancer Res. 1990,50,69716976. Ware, D. C.; Wilson,W. R.; Denny, W.A.;Rickard, C. E. F.Design and syntheaisof cobalt(III) nitrogen mustard complexesashypoxiaselective cytotoxins. The X-ray crystal ntructure of bis(3-chloro2,Cpentanedionato)RS-N,N'-bis(2-~o~thyl)ethylen~iamine)cobalt(1II) perchlorate, Co[(Clacac)~(BCE)]C104.J. Chem. SOC., Chem. Commun. 1991, 1171-1173. Price, C. C.; Kabas, G.; Nakata, I. Some amino and ammonio nitrogen mustard analogues. J. Med. Chem. 1965,8,650-655. Peck, R. M.; Preston, R. K.; Creech, H. J. Nitrogen mustard 1959,&?1,3984analoguesof antimalarial drugs. J.Am. Chem.SOC. 3989. Vargha,L.; Toldy, L.; Feher, 0.;Lendvai, S. Synthesisof new sugar derivatives of potential antitumour activity. Part I. Ethylenediamino- and 2-chloroethylamino-derivatives. J. Chem. SOC.1967, 805-812. Boucher, L. J.; Bailar, J. C. The preparation and properties of sodiumdinitro-bie-acetylacetonato-mbalt(III)and somecomplexes of the typenitro-.amminebi-acetylacetonato-cobalt(m). J.Znorg. N u l . Chem. 1966,27,1093-1099. Nakajima, K.; Kojima, M.; Fujita, J. Inversion at selenium in the b i ~ ~ t o ) - r ~ ( m e ~ y ~ o ion. ) eBull. t h y l SOC.Chem. Jpn. 1986,59,3505-3510. Bond, A. M.; Lawrance, G. A.; Lay, P. A.; Sargeson, A. M. Electrochemistry of macrocyclic(hexamine)cobalt(III) complexes. Metal-centered and substituent reductions. Znorg. Chem. 1986, 22,2010-2021. Jannakoudakie, A. D.; Tsiamie, C.; Jannakoudakis, P. D.; Theodoridou,E. Electrochemicalbehaviour of tris@diketonato)cobalt(III) chelates on mercury and carbon fibre electrodea. J. Electroanal. Chem. Interfacial Electrochem. 1985.184.123-133. (32) Thompson, L. H.; Rubin, J. S.; Cleaver, J. E.; Whitmore, G. F.; Brookman, K. A screening method for isolating DNA repairdeficient mutants of CHO cella. Somat. Cell. Genet. 1980,6,391405. (33) Hoy, C. A.;Thompson, L. H.; Mooney,C. L.; Salauu,E. P. Defective DNA croes-link removal in Chinese hamster cell mutants hypersensitive to bifunctional alkylating agents. Cancer Res. 1985,45, 1737-1743. (34) Wilson, W. R.; Thompson, L. H.; Anderson, R. F.; Denny, W. A. Hypoxia-selective antitumor agents. 2. Electronic effects of CsuMituents on the mechanisms of cytotoxicity and metabolic stability of nikacrine analogues. J. Med. Chem. 1989,32,31-38. (35) Whitmore, G. F.; Gulyas, S. Studies on the toxicity of RSU-1069. Znt. J. Radiat. Oncol. Biol. Phys. 1986,12,1219-1222. (36) Palmer, B. D.; Wileon,W. R.; Pullen, S. M.; Denny, W. A. Hypoxiaselective antitumor agents. 3. Relationships between structure and cytotoxicity against cultured tumor cella for substituted N,Nbis-(2-chloroethy1)anilines. J. Med. Chem. 1990,33,112-121. (37) Chaplin, D. J.; Acker, B.; Olive, P. L. Potentiation of the tumor cytotoxicity of melphalan by v a e d i t i n g agents. Znt. J. Radiat. Oncol. Biol. Phys. 1989,16,1131-1135. (38) In the prelhhary report (ref 24), the hypoxic selectivity of 21 was determined aa about &fold. These studies were carried out on material containing ca. 10%impuritiea The experiments reported here were carried out using material purified by reverse-phase HPLC; this consistently shows a hypoxic selectivity of 20-fold in the same assay. (39) Wilson,W. R.; Moselen, J. W.; Cliffe, S.; Denny, W. A,; Ware, D. C. Exploiting tumor hypoxia through bioreductive release of diffusible cytotoxins;the cobalt(III) nitrogen mustard complexSN 24771. Manuscript in preparation. (40)Chen, I. R. Zn situ detection of Mycoplusma contamination in cell culturea by fluorescent Hoechst 33258 stain. Exp. Cell Res. 1977, 104,255-262. (41) Finlay, G. J.; Wileon, W. R.; Baguley, B. C. A semiautomatad microculture method for investigating growth inhibitory effects of cytotoxic compounds on exponentially growing carcinoma cella. Anal. Biochem. 1986,139,272-277.